Rakhigarhi's Ancient DNA, Revisited: What the 2019 Cell Paper Did and Didn't Show

A single tooth, one skeleton, and four years of laboratory work produced one genome. That genome has since been asked to settle an argument that has run for over a century. It cannot, and the paper that published it never claimed it could.

The 2019 study, "An Ancient Harappan Genome Lacks Ancestry from Steppe Pastoralists or Iranian Farmers", appeared in Cell, authored by Vasant Shinde, Vagheesh Narasimhan, Niraj Rai, and a large international team including David Reich's Harvard lab. It reported a low-coverage genome from a skeleton excavated at Rakhigarhi, Haryana, a Harappan-phase site, dated by the excavators to roughly 2800 BCE. The team had processed dozens of skeletal samples from the site; only one yielded usable ancient DNA. That is the number that gets lost every time this study is summarized in a headline or a tweet.

What the genome actually contained

The individual's ancestry, per the paper, derived mainly from two older populations the authors label Ancient Ancestral South Indians (AASI, a hunter-gatherer-related lineage) and a group related to Iranian agriculturalists — but critically, a lineage that split from Iranian farmers before the Iranian farming population itself acquired West Asian ancestry through later admixture. The authors describe this combination as characteristic of what they call "Indus Periphery" individuals — a set of genomes from three individuals found not at Harappan core sites but at Gonur (Turkmenistan) and Shahr-i-Sokhta (Iran), sites on the Indus Civilization's periphery, published by the same team in an earlier companion paper, Narasimhan et al., "The Formation of Human Populations in South and Central Asia," Science, 2019.

The Rakhigarhi genome, the authors argued, showed no detectable Steppe pastoralist ancestry and no ancestry from post-migration Iranian farmer populations who had already mixed with Anatolian or Levantine groups. That absence is the paper's actual finding. It is a statement about one individual from one burial context at one site in one time period.

Where the citation habit breaks down

In public debate, this single result routinely gets stretched into two incompatible claims. One camp cites it to argue the Harappans were "pure" indigenous people untouched by any later migration, implying the entire Indo-Aryan migration model is wrong. The other camp cites the same paper's broader Central Asian dataset — which does document Steppe-related ancestry entering South Asia in the second millennium BCE, per the companion Science paper — to argue migration is settled fact for the region as a whole.

Both readings run past what a single ancient genome, from a Harappan-phase burial, can support. The Cell paper's authors themselves were explicit that their sample size was one individual, and that inferences about the wider Harappan population's genetic makeup rested on modeling this genome alongside the larger South Asian and Central Asian dataset from the companion study, not on Rakhigarhi alone. Vagheesh Narasimhan and coauthors framed the Rakhigarhi individual as consistent with the "Indus Periphery" cline already reconstructed from the Gonur and Shahr-i-Sokhta genomes — a supporting data point, not a foundational one.

The archaeology under the genetics

The genetic result cannot be separated from the excavation record it sits on. Vasant Shinde's team's excavation reports on Rakhigarhi, published through ASI channels and referenced in the Cell paper's methods, document cemetery contexts (RGR7.1) from which the skeletal material was drawn, with associated Harappan-phase pottery and burial goods. Any dating dispute about the individual's age is, at bottom, a dispute about that stratigraphic context — which layer the burial cut into, what sealed it, what could have intruded later. The Cell paper reports radiocarbon-based dating placing the sample in the range consistent with Mature Harappan occupation, but readers evaluating "how old" and "how representative" this genome is should treat those as two separate questions: one answered by radiocarbon and stratigraphy, the other requiring many more genomes than currently exist from Harappan cemetery contexts.

What would actually resolve the argument

The Rakhigarhi paper's own discussion section calls for more ancient DNA from Harappan sites before drawing population-level conclusions. That call has gone largely unanswered in the years since — ancient DNA preservation in South Asia's hot, humid soil chemistry is poor, which is why one usable sample came out of a much larger attempted set in the first place. Until multiple genomes from multiple Harappan-phase cemeteries are sequenced and published — ideally with full peer review of coverage depth and contamination controls, the standard Antiquity and Science reviewers apply to ancient DNA claims — the Rakhigarhi genome will keep being cited as if it were a verdict.

It isn't one. It is a single, carefully obtained, genuinely informative data point that fits into a larger and still incomplete picture built from Central Asian and Iranian plateau genomes. Treating it as more than that — in either direction — is asking one skeleton to do the work of a population.